xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-19 20:37:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

misc fixes

Browse Source 
Signed-off-by: Sage Moore <sage@neuralmagic.com>
This commit is contained in:
Sage Moore 2025-08-05 19:23:23 +00:00
parent 4819bb8715
commit 6b0c303ab4
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
vllm/compilation/decorators.py
View File

@ -189,7 +189,6 @@ def _support_torch_compile(
CompilationLevel.NO_COMPILATION, CompilationLevel.DYNAMO_AS_IS CompilationLevel.NO_COMPILATION, CompilationLevel.DYNAMO_AS_IS
] or not supports_dynamo() or _should_ignore_torch_compile( ] or not supports_dynamo() or _should_ignore_torch_compile(
self.__class__) self.__class__)
self.do_not_compile = True
if self.do_not_compile: if self.do_not_compile:
return return

3
vllm/v1/worker/gpu_model_runner.py
View File

@ -1918,11 +1918,12 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
scheduler_output, is_dummy_run) scheduler_output, is_dummy_run)
# if is_global_first_rank(): # if is_global_first_rank():
# logger.info(f"RUNNING FULL BATCH {num_scheduled_tokens}") # logger.info(f"RUNNING FULL BATCH {num_scheduled_tokens}")
skip_cuda_graphs = self.parallel_config.enable_microbatching
with set_forward_context(attn_metadata, with set_forward_context(attn_metadata,
vllm_config=self.vllm_config, vllm_config=self.vllm_config,
num_tokens=num_scheduled_tokens or 1, num_tokens=num_scheduled_tokens or 1,
num_tokens_across_dp=num_tokens_across_dp, num_tokens_across_dp=num_tokens_across_dp,
skip_cuda_graphs=True): skip_cuda_graphs=skip_cuda_graphs):
return self.model( return self.model(
input_ids=input_ids, input_ids=input_ids,
positions=positions, positions=positions,